Types of networks (LAN, WAN, MAN, PAN)

The two-host example from What is a computer network? is a network in the strict sense, but it doesn't look much like the network you actually use every day. Your phone talks to a Wi-Fi router a few meters away, that router talks to an ISP, and the ISP connects to servers that might be on another continent. Distance changes everything about how a network gets built — the cabling, the addressing scheme, who owns the equipment, even which protocols make sense.

Networking engineers group networks by the geographic scope they cover, because scope drives almost every design decision that follows.

Why scope matters

A network confined to one room can afford things a global network can't. If there are only twenty devices in the building, it's cheap for one device to send a message to literally everyone else and let them sort out who it was for — that's how a fair amount of local traffic works. Do the same thing across the entire internet and you'd flood every router on the planet with noise meant for one host in one city.

So as scope grows, networks trade the cheap, simple techniques that work at small scale for structured, hierarchical ones: routing instead of broadcasting, dedicated address blocks instead of ad hoc numbering, infrastructure owned by telecom carriers instead of one company's IT closet. The categories below track that tradeoff.

PAN — Personal Area Network

A Personal Area Network covers the space immediately around one person, typically a few meters. Bluetooth earbuds paired with a phone, a smartwatch syncing with that phone, a wireless keyboard talking to a laptop — all PANs.

PANs are short-range, low-power, and connect a handful of devices belonging to one person. Bluetooth and technologies like Zigbee, common in smart-home sensors, dominate this space because they're built to sip battery power rather than push throughput.

LAN — Local Area Network

A Local Area Network covers a single site: a home, an office floor, a school building. This is the network you interact with most directly — the Wi-Fi you connect to, the Ethernet jack in a server room.

A LAN is usually owned and administered by one organization or household, uses private IP address ranges internally (covered fully in Public vs Private IP), and relies on Ethernet or Wi-Fi as its physical layer. Because everyone on it is close together and under one administrative roof, a LAN can use fast, broadcast-friendly techniques that wouldn't scale further out. The clearest example is how a host finds the hardware address of a neighbour it wants to reach: it simply asks every device on the local segment at once and waits for the one that owns that address to answer. That works precisely because "every device on the segment" is a small, bounded set.

Most home networks and small offices are, technically, one LAN sitting behind a single router that connects it to the wider internet. That router is doing real work, not just passing traffic through — it sits on two networks at once and decides which traffic belongs on which side.

MAN — Metropolitan Area Network

A Metropolitan Area Network spans a city or a large campus — bigger than one building, smaller than a country. A university connecting several campuses across town, or a city government linking municipal offices, is running a MAN.

MANs sit in an awkward middle ground, and you'll hear the term less often today than LAN or WAN. Cheap, high-capacity fiber has made it easy for organizations to treat a metro-scale network as a handful of LANs stitched together over leased fiber, which blurs the distinction. Note what the blurring is really about: the technology and the ownership model, not the distance. A MAN typically runs over fiber leased from a carrier, whereas a LAN runs over cabling the organization installed and owns outright. That difference in who controls the physical layer is what makes it a separate category, and it's why "a MAN is just a big LAN" doesn't hold up.

WAN — Wide Area Network

A Wide Area Network spans a large geographic area — a country, a continent, or the entire planet. The internet itself is the largest WAN in existence: a network of networks, connecting millions of independently administered LANs together.

WANs are typically built and operated by telecommunications carriers and internet service providers, using long-haul fiber, undersea cables, and satellite links. No single organization owns "the internet" — it's a cooperative arrangement where networks agree to exchange traffic with each other, coordinated by BGP, the protocol by which each network announces to its neighbours which blocks of addresses it can deliver traffic to.

A corporate WAN is a smaller, private version of the same idea: a company with offices in Tashkent, Istanbul, and Berlin might connect those three LANs using a VPN — an encrypted tunnel across the public internet that makes distant networks behave as though they were directly wired together — or a dedicated leased line. Either way the result is one logical WAN under a single company's control. That's also why "WAN" and "the internet" aren't synonyms: a private corporate WAN can ride entirely on public internet cabling and still be a separate, private network.

Putting the scales side by side

Don't treat the boundaries as precise measurements — there's no rule that a network becomes a MAN at exactly five kilometers. What actually distinguishes these categories in practice is administrative and technical: who owns the equipment, whether private or public addressing is used, and whether the network leans on local switching or internet-scale routing.

Practical scenario: choosing the right scope

Imagine a small backend team has three environments:

Environment        Network scope that fits
-----------------  -----------------------------------------------------
Developer laptop   PAN/LAN: local containers, phone tethering, Wi-Fi
Office servers     LAN: internal services, printers, shared storage
Cloud production   WAN: users and services spread across regions

The difference is not cosmetic. A database exposed only on an office LAN can rely on private addressing and strict internal access rules. The same database exposed across a WAN needs encryption in transit, careful firewall rules, monitoring, and a plan for latency. A file copy across a LAN may feel instant; the same copy between cloud regions crosses provider backbones and can be limited by distance, congestion, and egress policy.

When someone says "the network is slow," ask which network. A slow Bluetooth headset, a crowded office Wi-Fi LAN, and a cross-region cloud request are three different problems wearing the same vague label.

Practical scenario: the LAN works, the WAN doesn't

A new server gets racked, cabled, and given an address by hand. Every other machine on the same office LAN can reach it — file shares, internal dashboards, SSH, all fine. It just can't reach anything outside the building: no package updates, no external API calls, nothing.

Start with the interface itself, since a LAN and a WAN are two different scopes and the fault could sit in either:

ip addr show eth0

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
    link/ether 08:00:27:1a:2c:9f brd ff:ff:ff:ff:ff:ff
    inet 10.0.5.40/24 brd 10.0.5.255 scope global eth0

The interface is up, the address is valid, and ping 10.0.5.1 (a colleague's desk machine, same /24) works — so the LAN scope is fine. The next question belongs to the WAN scope: does this host even know how to leave its own network?

ip route show

10.0.5.0/24 dev eth0 proto kernel scope link src 10.0.5.40

That's the entire routing table. There's no line starting with default via — no default route, no gateway to hand traffic to once the destination isn't inside 10.0.5.0/24. The kernel has a perfectly good local map and nothing beyond it. ping 10.0.5.1 never needed a gateway; ping 8.8.8.8 needs one and the kernel has nowhere to send it, so it fails immediately with a routing error rather than a timeout.

sudo ip route add default via 10.0.5.1 dev eth0

That command is not persistent across a reboot on most distributions — it edits the live routing table, not a config file — which is deliberate for testing. Once confirmed working, the equivalent needs to go into the distribution's actual network configuration (Netplan, systemd-networkd, or /etc/network/interfaces, depending on the system) so it survives a restart. Verify with ip route show again and a fresh ping 8.8.8.8.

This is the practical difference between LAN and WAN scope in one incident: a LAN only needs hosts to agree on a shared address range and use ARP to find each other directly. Reaching anything outside that range needs one more piece of information entirely — a gateway — and no amount of correct LAN addressing substitutes for it.

What a strong answer should include

If asked to explain why an office network can be "up" internally while having no internet access, a strong answer names the missing piece precisely: a default route (or gateway) rather than a vague "something's misconfigured." It should also state the distinguishing test — pinging an address inside the local subnet succeeds using only ARP and local delivery, while pinging anything outside it requires the routing table to name a next hop. Bonus points for noting that "no internet" and "DNS not resolving" are different failures with different first commands (ip route show versus dig), even though both look identical to an end user staring at a browser error.

Common mistakes

• Reading the categories as distance thresholds. They describe ownership and routing model first, geography second. A campus network and a two-building network can both be LANs if one organization owns the cabling between them.
• Assuming scope doesn't affect application design. It does, directly. Retry logic, timeout values, and whether you can treat a call as "basically instant" all change depending on whether the other end is in the same rack or another continent.

Practice exercises

1. Classify each of the following as PAN, LAN, MAN, or WAN, and justify your answer: a fitness tracker syncing to a phone; a five-branch bank connecting all its offices in one country; a university's Wi-Fi covering three adjacent buildings; two friends video-calling across different countries.
2. Your home router connects your laptop, phone, and smart TV to your ISP. Identify which side of the router is the LAN and which side is the WAN connection, and explain how you could confirm this from the device list in your router's admin page.
3. Check whether your own internet service provider operates its own long-haul fiber or leases capacity from a larger carrier. What does that tell you about how many separate networks your traffic actually crosses before reaching a website hosted abroad?
4. The "ask every device on the segment at once" technique described in the LAN section is reasonable on a LAN and unthinkable on the internet. Work out roughly how much traffic it would generate if a single such request had to reach every host on the internet, and use that to explain in your own words why scope forces a different design.

Exercise 4 is the hinge for the next article. Broadcasting to "everyone on the segment" only makes sense if you know what the segment physically looks like — how many devices share it, and what happens when one cable in the middle fails. That shape has a name, and it has consequences for reliability that show up long before scale does: Network topologies is next.

Sources

• Cisco, What Is a LAN? Local Area Network Definition
• Cloudflare Learning Center, What is a WAN? | Wide area network
• IETF, RFC 1918 – Address Allocation for Private Internets — the private addressing scheme most LANs rely on internally.